On Fri, Dec 25, 2009 at 3:10 PM, Robert Haas <robertmhaas@gmail.com> wrote:
> On Fri, Dec 25, 2009 at 2:30 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> Robert Haas <robertmhaas@gmail.com> writes:
>>> Works for me. One small problem discussed upthread is that there
>>> currently doesn't appear to be a libpq function that does
>>> ident-quoting. I'm thinking that we will need to add one to make this
>>> work - is that going to be a problem?
>>
>> The rules for ident-quoting are simple and haven't changed over the
>> years, so we don't really *need* a libpq function for it. OTOH you
>> could argue it's inconsistent that we have one and not the other.
>
> Yeah. Plus it seems like a useful thing to have, anyway.
>
>>> I'm thinking that since we're
>>> just adding a function it won't force an uncomfortable major-version
>>> bump on libpq.
>>
>> Yeah, we have taken the position in the past that adding new functions
>> doesn't require a soname bump.
>
> Good.
So it seems we have agreement on a new direction for this work. We
will not add the \pexec option Pavel proposed as part of this patch;
instead, we will consider a patch that makes :'foo' and :"foo" do
literal and identifier quoting of the corresponding value. Based on
this, I am marking the existing patch as Returned with Feedback, since
what is needed here will amount to a totally base of code, and we can
consider the revised patch if any for whichever CommitFest is open at
the time that patch is submitted.
Thanks,
...Robert